Table 11.

The performance of different classifiers before and after matching.

Status		Recall	Specificity	Accuracy	Precision
NN-SGDa
	Before matching	21.95	99.46	98.98	20.45
	After matching	51.22	97.75	97.46	12.50
NN-Adamb
	Before matching	70.73	97.32	97.16	14.22
	After matching	58.54	96.45	96.22	9.38
WMVc (NN-SGD)
	Before matching	85.37	41.31	58.85	1.28
	After matching	100.00	73.04	73.20	2.27
WMV (NN-Adam)
	Before matching	92.68	90.20	90.21	5.60
	After matching	85.37	89.41	89.38	4.81

^aNN-SGD: neural network model using stochastic gradient descent.

^bNN-Adam: neural network model using Adam.

^cWMV: weighted majority voting.